OpenCode Configuration Guide: Connect to the NextHub API
OpenCode is a terminal-based AI coding tool that supports multiple models. Below are the steps for connecting to NextHub's Claude channel.
Setup Steps
1. Install OpenCode
Make sure Node.js is installed, then install OpenCode globally:
npm i -g opencode-aiAfter installation, run it once so it generates the default configuration folder:
opencode2. Open the OpenCode configuration folder
start "" "%USERPROFILE%\.config\opencode"open "$HOME/.config/opencode"3. Edit opencode.json — Add the Claude configuration
Find opencode.json in the configuration folder (create it if it doesn't exist), and fill in your ApiKey and the NextHub endpoint under provider.anthropic.options following the official OpenCode schema:
{
"$schema": "https://opencode.ai/config.json",
"provider": {
"anthropic": {
"options": {
"apiKey": "your NextHub ApiKey",
"baseURL": "https://api.nexthub.cc"
}
}
}
}Don't add /v1 to the Base URL
OpenCode automatically appends /v1/messages to the baseURL, so just use the bare domain https://api.nexthub.cc here — don't add /v1 at the end.
4. Verify the configuration
Run opencode, enter the /models command, and you're set if Claude models show up in the list:
opencode
# run after entering the interactive interface
# /modelsConnecting to the GPT Channel as Well
Add an openai provider to the same opencode.json to use the GPT / Codex model families:
{
"$schema": "https://opencode.ai/config.json",
"provider": {
"anthropic": {
"options": {
"apiKey": "your NextHub ApiKey",
"baseURL": "https://api.nexthub.cc"
}
},
"openai": {
"options": {
"apiKey": "your NextHub ApiKey",
"baseURL": "https://api.nexthub.cc/v1"
}
}
}
}The Two Base URLs Follow Different Conventions
anthropic uses the Anthropic protocol, so baseURL takes the bare domain https://api.nexthub.cc (/v1/messages is appended automatically); openai uses the OpenAI-compatible protocol, so baseURL must include /v1. Choose the matching group for each of the two ApiKeys (a Claude group and the GPT-Azure / GPT-Pro group). You can also use the same key for both — but the group determines which models are available, so configuring them separately is usually clearer.